January 22, 2015
1 min read
Save

Bone marrow lesion identification reliable with computer-assisted segmentation in knee OA

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Bone marrow lesions in the knees of patients with osteoarthritis were identified reliably using computer-assisted segmentation, which was sensitive to changes over time and less time-consuming compared with manual segmentation procedures.

Of 83 patients with knee osteoarthritis (OA) within the medial femoro-tibial compartment who had MRI performed at baseline and follow-up, researchers randomly selected 22 for further evaluation. The median interval between baseline and follow-up MRI was 334 days.

MRI was used to obtain fat-saturated/short tau inversion recovery (STIR), T1 and fat-saturated T-1 post-contrast sequences, and readings were assessed independently. Using the signal intensities of normal bone barrow in the lateral femoral condyles and tibial plateaus as threshold values, the researchers assessed bone marrow lesion (BML) volume medially through the use of manual segmentation and computer-assisted segmentation.

Results showed computer-assisted and manual segmentation obtained nearly identical mean threshold values at 182.1 and 181.6, respectively. Agreement between inter- and intraobservers for the reference threshold values was high, according to the researchers.

Although the BML volumes obtained through computer-assisted and manual segmentation were significantly correlated, the researchers found the tissue changes measured were different. Additionally, computer-assisted segmentation measured the volume of voxels that exceeded the threshold values, whereas manual segmentation included intervening voxels with normal signal intensities.

Processing time for manual segmentation averaged between 30 and 45 minutes, compared with 15 minutes using computer-assisted segmentation, which also calculated pixel volume precisely, according to the researchers. – by Shirley Pulawski

Disclosure: The authors have no relevant financial disclosures.